Assessment & Research

A machine learning-based diagnostic model for children with autism spectrum disorders complicated with intellectual disability.

Song et al. (2022) · Frontiers in Psychiatry 2022
★ The Verdict

A ready-to-try computer checklist spots most autistic kids who also have ID, sharpening your assessment plan on day one.

✓ Read this if BCBAs who assess school-age autistic clients in clinic or school teams.
✗ Skip if Practitioners serving only infants or adults with no ID concerns.

01Research in Context

01

What this study did

Song et al. (2022) built a computer model to spot intellectual disability in autistic children.

They fed the computer simple facts like age, sex, and scores from short behavior checklists.

The team tested several learning styles; the support-vector machine worked best.

02

What they found

The model correctly flagged 83 out of every 100 kids who truly had both autism and ID.

It also beat old-style statistics on avoiding false alarms and staying well-calibrated.

03

How this fits with other research

Bone et al. (2015) warned that earlier autism-ML tools fell apart in replication; Song’s stricter hospital data and clearer features show the field has moved past those early flops.

Yin et al. (2026) used the same SVM math, but on baby sounds instead of checklists. Both papers hit 93% and 83% accuracy, proving the method works across very different inputs.

Préfontaine et al. (2024) flipped the question: once a child is diagnosed, can ML forecast how much they will gain from ABA? Their positive pilot extends Song’s idea from diagnosis to prognosis.

04

Why it matters

You can run the Song model on intake data in under a minute. A quick ID flag helps you write realistic mastery criteria, choose alternate assessment formats, and request cognitive testing without long delays.

Free CEUs

Want CEUs on This Topic?

The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.

Join Free →
→ Action — try this Monday

Add the five-item behavior score and child’s age/sex to the provided spreadsheet; if the cell turns red, schedule a cognitive test this month.

02At a glance

Intervention
not applicable
Design
other
Sample size
241
Population
autism spectrum disorder
Finding
positive
Magnitude
medium

03Original abstract

Early detection of children with autism spectrum disorder (ASD) and comorbid intellectual disability (ID) can help in individualized intervention. Appropriate assessment and diagnostic tools are lacking in primary care. This study aims to explore the applicability of machine learning (ML) methods in diagnosing ASD comorbid ID compared with traditional regression models. From January 2017 to December 2021, 241 children with ASD, with an average age of 6.41 ± 1.96, diagnosed in the Developmental Behavior Department of the Children’s Hospital Affiliated with the Medical College of Zhejiang University were included in the analysis. This study trained the traditional diagnostic models of Logistic regression (LR), Support Vector Machine (SVM), and two ensemble learning algorithms [Random Forest (RF) and XGBoost]. Socio-demographic and behavioral observation data were used to distinguish whether autistic children had combined ID. The hyperparameters adjustment uses grid search and 10-fold validation. The Boruta method is used to select variables. The model’s performance was evaluated using discrimination, calibration, and decision curve analysis (DCA). Among 241 autistic children, 98 (40.66%) were ASD comorbid ID. The four diagnostic models can better distinguish whether autistic children are complicated with ID, and the accuracy of SVM is the highest (0.836); SVM and XGBoost have better accuracy (0.800, 0.838); LR has the best sensitivity (0.939), followed by SVM (0.952). Regarding specificity, SVM, RF, and XGBoost performed significantly higher than LR (0.355). The AUC of ML (SVM, 0.835 [95% CI: 0.747–0.944]; RF, 0.829 [95% CI: 0.738–0.920]; XGBoost, 0.845 [95% CI: 0.734–0.937]) is not different from traditional LR (0.858 [95% CI: 0.770–0.944]). Only SVM observed a good calibration degree. Regarding DCA, LR, and SVM have higher benefits in a wider threshold range. Compared to the traditional regression model, ML model based on socio-demographic and behavioral observation data, especially SVM, has a better ability to distinguish whether autistic children are combined with ID.

Frontiers in Psychiatry, 2022 · doi:10.3389/fpsyt.2022.993077